Objectively measured sleep and body mass index: a prospective bidirectional study in middle-aged and older adults

Background: In recent years, short sleep has been increasingly recognized as a risk factor for obesity. However, current evidence has so far been limited to cross-sectional studies or longitudinal studies using self-reported sleep. Therefore, we explored the directionality of the association between objectively measured sleep and body mass index (BMI). Methods: The study consists of 1031 participants from the general population (52% women, 45-91 years at baseline). Sleep, BMI and waist circumference (WC) were measured twice across a follow-up of six years. BMI and WC were measured at the research center. Total sleep time (TST, hrs), sleep onset latency (SOL, min), sleep efficiency (SE, %) and wake after sleep onset (WASO, min) were estimated by a wrist-worn actigraph. In addition, cross-sectional and longitudinal associations in both directions were explored. Results: An hour shorter TST was cross-sectionally associated with approximately 0.5 kg/m(2) higher BMI. Longitudinally, longer TST and higher SE were associated with lower BMI (beta(TST) = -0.75, 95% CI: -1.08, -0.42; beta(SE) = -0.04, 95% CI: -0.08, -0.01). Conversely, one kg/m(2) higher BMI was prospectively associated with 0.02 h shorter TST (95% CI: -0.03, -0.01), and this association was more pronounced over time. Results from analyses with WC were in line with those of BMI. Conclusions: This is the first study to explore bidirectionality in the association between objectively measured sleep and BMI in a large population of middle-aged and older adults. Indices of poor sleep were associated with higher and less stable BMI across time. Conversely, a high BMI was associated with a decrease in sleep duration. This confirms that the relation between sleep and body size is bidirectional, and changes in either sleep or BMI are likely to co-occur with changes in health through multiple pathways. (C) 2019 Elsevier B.V. All rights reserved.